Algorithm for Detecting Anomalous Student Activities in the Online Learning Process Based on Box Plots

The proposed research presents an algorithm for detecting anomalous student activities during online Python programming education, based on box plot analysis. Throughout their learning journey, students submit solutions to unique exercises across a range of tasks for assessment within the Digital Teacher Assistant (DTA) system. Anomalous activities are defined as those involving an excessively high number of solution submissions for a particular task or those characterized by extremely prolonged durations of task-solving. In both cases, students require timely assistance from instructors to clarify challenging course material. Upon each new submission of an exercise solution into the DTA system, the characteristics of the box plots for that specific task are recalculated. Decisions are made regarding whether the student's submission constitutes an anomaly (outlier), either in terms of the number of submissions or the duration of task-solving. The experimental results, using a dataset collected during the spring semester of 2023 at RTU MIREA, indicate the practical viability of implementing the proposed algorithm for the real-time detection of various anomalous activities (outliers) in students' interactions with the DTA system.